www.bba-direct.com

Biochimica et Biophysica Acta 1604 (2003) 55–59

Rapid report

Properties of the apo-form of the NADP(H)-binding domain III of

proton-pumping Escherichia coli transhydrogenase: implications

for the reaction mechanism of the intact enzyme

Anders Pedersen1, Jenny Karlsson1, Magnus Althage, Jan Rydstrom*

Department of Biochemistry and Biophysics, Goteborg University, Box 462, 405 30, Goteborg, Sweden

Received 6 March 2003; received in revised form 14 March 2003; accepted 25 March 2003

Abstract

Proton-translocating nicotinamide nucleotide transhydrogenases contain an NAD(H)-binding domain (dI), an NADP(H)-binding domain

(dIII) and a membrane domain (dII) with the proton channel. Separately expressed and isolated dIII contains tightly bound NADP(H),

predominantly in the oxidized form, possibly representing a so-called ‘‘occluded’’ intermediary state of the reaction cycle of the intact

enzyme. Despite a Kd in the micromolar to nanomolar range, this NADP(H) exchanges significantly with the bulk medium. Dissociated

NADP+ is thus accessible to added enzymes, such as NADP-isocitrate dehydrogenase, and can be reduced to NADPH. In the present

investigation, dissociated NADP(H) was digested with alkaline phosphatase, removing the 2V-phosphate and generating NAD(H).

Surprisingly, in the presence of dI, the resulting NADP(H)-free dIII catalyzed a rapid reduction of 3-acetylpyridine-NAD+ by NADH,

indicating that 3-acetylpyridine-NAD+ and/or NADH interacts unspecifically with the NADP(H)-binding site. The corresponding reaction in

the intact enzyme is not associated with proton pumping. It is concluded that there is a 2V-phosphate-binding region in dIII that controls tight

binding of NADP(H) to dIII, which is not a required for fast hydride transfer. It is likely that this region is the Lys424–Arg425–Ser426

sequence and loops D and E. Further, in the intact enzyme, it is proposed that the same region/loops may be involved in the regulation of

NADP(H) binding by an electrochemical proton gradent.

D 2003 Elsevier Science B.V. All rights reserved.

Keywords: Transhydrogenase; NAD; NADP; Proton pump; Membrane protein

Nicotinamide nucleotide transhydrogenase (TH) from

Escherichia coli is a proton pump in which the reduction of

NADP+ by NADH through a hydride ion is linked to the

translocation of one proton from the periplasmic space to the

cytosol in bacteria, according to the reaction

Hþout þ NADHþ NADPþWHþ

in þ NADþ þ NADPH ð1Þ

in which the number of protons translocated per NADPH

generated is 1. In the presence of an electrochemical proton

0005-2728/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserv

doi:10.1016/S0005-2728(03)00028-8

Abbreviations: TH, transhydrogenase; dI, domain I of transhydroge-

nase; dII, domain II of transhydrogenase; dIII, domain III of trans-

hydrogenase; ecI–ecIII, the corresponding domains of Escherichia coli

transhydrogenase; rrI, dI of Rhodospirillum rubrum transhydrogenase;

AcPyAD+, 3-acetylpyridine-NAD+; NADP-ICDH, NADP(H)-specific iso-

citrate dehydrogenase; ADH, alcohol dehydrogenase

* Corresponding author. Tel.: +46-31-7733921; fax: +46-31-7733910.

E-mail address: jan.rydstrom@bcbp.gu.se (J. Rydstrom).1 These authors contributed equally to this work.

gradient, Dp, reaction (1) from left to right is activated some

5–10-fold and the apparent equilibrium is strongly shifted to

the right.

The E. coli enzyme is composed of an a subunit (54.6

kD) and a h subunit (48 kD), containing the NAD(H)-

binding domain I (dI) and the NADP(H)-binding domain III

(dIII), respectively. Intact and active TH is a tetramer, a2h2.

Both subunits also contain transmembrane helices, of which

helices 1–4 reside in the a subunit and helices 6–14 reside

in the h subunit, together constituting the membrane domain

(dII) which also houses the proton channel (Fig. 1). Phys-

iologically, the role of TH is presumably to generate a high

redox level of NADP(H) for detoxification and regulatory

purposes (for reviews, see Refs. [1,2]). Despite the fact that

both dI [3] and dIII [4,5] as well as the dI–dIII complex [6]

have been structurally resolved by X-ray crystallography,

and that dI–dIII interactions have been established by NMR

[7,8], both approaches providing essential insights into the

hydride transfer mechanism, the coupling mechanism of

ed.

Fig. 1. A cartoon of the E. coli transhydrogenase and its domains, shown as the ah-monomer.

A. Pedersen et al. / Biochimica et Biophysica Acta 1604 (2003) 55–5956

intact transhydrogenase is unknown. Even though Dp-

dependent alterations cannot be excluded in dI, the general

view is that dissociation of NADP(H) from dIII constitutes

the limiting factor in both directions of reaction [1] includ-

ing the Dp-stimulated forward reaction (left to right) [1,2].

One essential observation in this context is the fact that

isolated dIII contains tightly bound NADP(H), assumed to

represent an ‘‘occluded’’ state in the overall reaction mech-

anism [2]. Thus, in the presence of isolated dI and sub-

strates, dIII catalyzes very slow forward and reverse

reactions, but a high rate of cyclic reduction of 3-acetylpyr-

idine-NAD+ by NADH, mediated by the bound NADP(H).

An important question concerns what the structural differ-

ences are between dIII and the corresponding domain in the

intact enzyme that has no bound NADP(H). To this end we

have generated the apo-form of dIII from E. coli (ecIII) and

characterized its properties. Fig. 2 shows a close-up of the

NADP(H)-binding site in dIII.

EcIII was expressed and purified as described [7,9].

NADP+ release from ecIII was monitored by fluorescence

as described previously [9]. All measurements were made

with a SPEX model FL1T1 t2 spectrofluorometer with both

excitation and emission slits set to 2.5 nm. The excitation and

emission wavelengths used were 340 and 460 nm, respec-

tively. Alkaline phosphatase from bovine intestinal mucosa

was used to remove the 2V-phosphate from NADP(H). Each

measurement was preceded by incubation of 120 Ag of ecIII

with 20 U alkaline phosphatase in 10 mM Tris–HCl, pH 9, at

37 jC for 30 min in a total volume of 100 Al. Prior to activitymeasurement, the incubation mix was diluted to 2 ml with 20

mMMOPS, 5 mMMgCl2, pH 7.0. NADP+ was assayed with

isocitrate dehydrogenase in the presence of isocitrate and

MgCl2 [10] and NADH formation was assayed using alcohol

dehydrogenase (ADH) in the presence of semicarbazide and

ethanol according to [10]. Cyclic activities were measured

optically using Rhodospirillum rubrum domain I (rrI) as

described [9] in a medium composed of 20 mM CHES, 20

mMMES, 20 mMOPS, 20 mM Tris, 50 mM NaCl (pH 7.0).

Incubations of 18 Ag ecIII in 10 mM Tris–HCl, pH 9, at 37

jC for 10 min either with or without 10 U of alkaline

phosphatase were made prior to activity measurements.

The total incubation volume was always 50 Al. Due to its

stabilising effect on activity, 300 AM NADH was added

during the incubation.

Fig. 2. The NADP(H)-binding site in the dIII crystal structure of the bovine mitochondrial TH (PDB code 1D4O). Bound NADP(H) is shown as a ball and

stick structure with the nicotinamide moiety to the left and the adenine moiety to the right. Residues Lys424–Arg425–Ser426 binding the 2V-phosphate of

NADP(H), residues Val347 and Tyr431 binding the nicotinamide ring, and residues Arg350 and Asp392 binding the pyrophosphate moiety of NADP(H),

are indicated and numbered according to the equivalent E. coli residues. Loops D and E are also indicated. The figure was created with the software

MOLMOL [19].

A. Pedersen et al. / Biochimica et Biophysica Acta 1604 (2003) 55–59 57

Wild-type ecIII contains about 8% apo-form and 92%

NADP(H), the latter distributed as 87% NADP+ and 5%

NADPH [9]. Despite the fact that NADP+ and NADPH are

bound to dIII with dissociation constants in the micromolar

and nanomolar range, respectively [7,9,11], both exchange

sufficiently fast to react with externally added reducing and

oxidizing enzymes [11]. This is exemplified in Fig. 3A,

where bound NADP+ is reduced by NADP-isocitrate dehy-

drogenase (NADP-ICDH) at a rate which is related to Kd,

whereas, as expected, the NAD-alcohol dehydrogenase in

the presence of semicarbazide and ethanol had no effect.

Because of the significant dissociation rate for NADP(H), it

was assumed that the 2V-phosphate of NADP(H) could be

hydrolyzed by added alkaline phosphatase, thereby remov-

ing NADP(H) from the binding site. The latter enzyme is

known to hydrolyze the 2V-phosphate of NADP(H) [12].

Indeed, following incubation of ecIII with alkaline phospha-

tase at pH 9.0 for 30 min and readjustment of the pH to 7.0,

NADP+ was no longer detectable by the NADP-ICDH assay

(Fig. 3C). Instead, ADH gave a fluorescence change, indi-

cating that NADP+ had been converted to NAD+ which then

was reduced. Shorter times of incubation at pH 9.0 gave a

larger amount of NADH (not shown), indicating that the

small amounts of NAD+ formed at pH 9.0 were unstable, in

agreement with the known instability of NAD(P)+ at high

pH. Added free NAD+ gave the expected additional increase

in absorbance due to the NADH formed (Fig. 3C). That the

incubation and pH shift in themselves had little or no effect

on the NADP+ content or apparent Kd is shown in Fig. 3B.

Bound NADPH dissociates some 50–100-fold slower

from ecIII than NADP+ [9]. Following phosphatase treat-

ment, the low amount of NADPH bound to ecIII was no

longer detectable enzymatically by, e.g. glutathione reduc-

tase in the presence of oxidized glutathione, suggesting that

it too had been degraded (not shown).

The rates of cyclic transhydrogenation catalyzed by

untreated and phosphatase-treated ecIII in the presence of

rrI and NADP+ were comparable to previous assessments of

rrI–ecIII Vmax, [9,11], i.e., about 5000 mol AcPyADH mol

ecIII� 1 min� 1 (Fig. 4, traces A–C). The remarkably high

activity of phosphatase-treated, and thus NADP(H)-free,

ecIII in the presence of rrI but in the absence of added

NADP+ (Fig. 4, trace D) demonstrates that ecIII was capable

of catalyzing an unspecific cyclic transhydrogenation

between NADH and AcPyAD+ at approximately 75% of

Vmax, corresponding to 3770 mol AcPyADH mol ecIII� 1

min� 1. The pH dependence of the phosphatase-treated ecIII

in the presence of rrI was not significantly different from that

of nontreated ecIII [13] (not shown).

Isolation of stable apo-forms of wild-type ecIII has not

been reported previously, the main problem being instability

and aggregation (M. Althage, O. Fjellstrom and J. Rydstrom,

unpublished). However, two mutants, D392C [7] and R425C

[14], showed 100% apo form as well as high reverse and low

cyclic activities, consistent with a markedly decreased bind-

ing of NADP(H). Indeed, this was expected since Asp392 is

essential for activity in intact TH [15]. Asp392 forms H-

bonds with the pyrophosphate moiety of NADP(H), and

Fig. 3. Removal of 2V-phosphate of NADP+ bound to ecIII with alkaline

phosphatase. (A) Reduction of NADP+ bound to untreated ecIII by NADP-

ICDH. (B) Conditions were as in (A) except that ecIII was exposed to pH

9.0 followed by a readjustment to pH 7.0. In both (A) and (B), ADH was

added to reduce any NAD+ formed. (C) Reduction of NADP+ and NAD+

bound to phosphatase-treated ecIII by NADP-ICDH and ADH, respec-

tively; NAD+ was subsequently added. Reduction of NADP+ and NAD+

was followed fluorimetrically. The concentrations of ethanol, semicarbazide

and isocitrate were 0.1%, 50 mM and 2 mM, respectively. The amounts of

isocitrate dehydrogenase (ICDH) and alcohol dehydrogenase (ADH) added

in each measurement were 1 and 10 U, respectively.

Fig. 4. Cyclic activities of untreated (A and C) and phosphatase-treated (B

and D) ecIII. For incubation conditions, see Materials and methods. The

concentrations of ecIII and rrI were 12.5 nM and 1 AM, respectively. The

concentrations of AcPyAD+ and NADH were 400 and 300 AM,

respectively. Measurements were carried out either in the presence (A

and B) or in the absence (C and D) of 200 AM NADP+. Additions were

made in the same order, i.e., ecIII, rrI, NADP+, AcPyAD+ and NADH.

A. Pedersen et al. / Biochimica et Biophysica Acta 1604 (2003) 55–5958

Arg425 forms H-bonds with the 2V-phosphate group of

NADP(H) [4,5].

That phosphatase-treated ecIII devoid of bound NADP(H)

showed a high cyclic activity with only AcPyAD+ and

NADH indicates that, in the absence of NADP(H), AcPyAD+

and/or NADH interacts unspecifically with the NADP(H)-

binding site in dIII. This observation is of profound impor-

tance for the mechanism as to how binding of NADP(H) may

be regulated. First, an unspecific interaction of dIII with

NAD(H) is sufficient for proper orientation of the nicotina-

mide ring of NAD(H) in dIII relative to that in dI, and fast

hydride transfer. Thus, the Lys424–Arg425–Ser426 region

that normally stabilizes the 2V-phosphate through H-bonds is

not required for fast hydride transfer. Second, the 2V-phos-phate group of NADP(H) is responsible for the tight binding

to dIII, most likely via the Lys424–Arg425–Ser426 region,

and indirectly by loops D and E (Fig. 2). The closing effect of

the latter loops is thus assumed to contribute to NADP(H)

binding only after the interaction between the 2V-phosphateand the Lys424–Arg425–Ser426 region has been estab-

lished. Third, it may be inferred that the Dp-dependent

increase in the dissociation of NADPH from the intact TH

involves a decreased binding of the 2V-phosphate of

NADP(H) to Lys424–Arg425–Ser426, as well as a concom-

itant increased competition with bulk NAD(H), i.e. a specif-

icity-change mechanism. The communication between the

protonation events in dII and the Lys424–Arg425–Ser426

region in dIII remains to be established.

A cyclic activity between AcPyAD+ and NADH cata-

lyzed by intact TH [16], cysteine-free TH [17] and some

mutants of the ‘‘hinge’’ region, e.g. R265C and S266C [18],

A. Pedersen et al. / Biochimica et Biophysica Acta 1604 (2003) 55–59 59

has indeed been reported previously that involves an unspe-

cific interaction of AcPyAD+ and/or NADH with the

NADP(H)-binding site. In addition, transfer of hydride ions

between AcPyAD+ and NADH in the absence of NADP+

does not involve proton pumping [16], suggesting that it is

indeed a cyclic reaction or, alternatively, that it is a ‘‘reverse’’

reaction where NADH has replaced NADPH in reaction (1).

Unfortunately, it is presently not possible to distinguish be-

tween a cyclic reaction and a reverse reaction in this context.

In conclusion, tightly bound NADP(H) in ecIII has been

successfully removed by treatment with alkaline phospha-

tase, producing the apo-form of ecIII that binds AcPyAD+/

NADH unspecifically but with a retained high activity for a

cyclic reaction. This suggests that primarily the 2V-phos-phate-binding Lys424–Arg425–Ser426 region, and secon-

darily loops D and E, are involved in the ‘‘occluded’’ state,

as well as the Dp-dependent binding changes in the

NADP(H) site of the intact enzyme.

Acknowledgements

This work was supported by the Swedish Science

Council. A.P. and J.K. acknowledge the support of the

Sven and Lilly Lawski Foundation.

References

[1] T. Bizouarn, O. Fjellstrom, J. Meuller, M. Axelsson, A. Bergkvist, C.

Johansson, B.G. Karlsson, J. Rydstrom, Proton translocating nicoti-

namide nucleotide transhydrogenase from E. coli. Mechanism of ac-

tion deduced from its structural and catalytic properties, Biochim.

Biophys. Acta 1457 (2000) 211–228.

[2] J.B. Jackson, S.A. White, P.G. Quirk, J.D. Venning, The alternating

site, binding change mechanism for proton translocation by transhy-

drogenase, Biochemistry 41 (2002) 4173–4185.

[3] P.A. Buckley, J.B. Jackson, T. Schneider, S.A. White, D.W. Rice, P.J.

Baker, Protein–protein recognition, hydride transfer and proton pump-

ing in the transhydrogenase complex, Structure 8 (2000) 809–815.

[4] S.A. White, S.J. Peake, S. McSweeney, G. Leonard, N.P.J. Cotton,

J.B. Jackson, The high resolution structure of the NADP(H)-binding

component (dIII) of proton-translocating transhydrogenase from hu-

man heart mitochondria, Structure 8 (1999) 1–12.

[5] G.S. Prasad, V. Sridhar, M. Yamaguchi, Y. Hatefi, C.D. Stout, Crystal

structure of transhydrogenase domain III at 1.2 A resolution, Nat.

Struct. Biol. 6 (1999) 1126–1131.

[6] N.J.P. Cotton, S.A. White, S.J. Peake, S. McSweeney, J.B. Jackson,

The crystal structure of an asymmetric complex of the two nucleotide

binding components of proton-translocating transhydrogenase, Struc-

ture 9 (2001) 165–176.

[7] A. Bergkvist, C. Johansson, T. Johansson, J. Rydstrom, B.G. Karls-

son, Interactions of the NADP(H)-binding domain III of proton-trans-

locating transhydrogenase from Escherichia coli with NADP(H) and

the NAD(H)-binding domain I studied by NMR and site-directed

mutagenesis, Biochemistry 39 (2000) 12595–12605.

[8] M. Jeeves, K.J. Smith, P.G. Quirk, N.P.J. Cotton, J.B. Jackson, Sol-

ution structure of the NADP(H)-binding component (dIII) of proton-

translocating transhydrogenase from Rhodospirillum rubrum, Bio-

chim. Biophys. Acta 1459 (2000) 248–257.

[9] O. Fjellstrom, C. Johansson, J. Rydstrom, Structural and catalytic

properties of the expressed and purified NAD(H)- and NADP(H)-

binding domains of proton-pumping transhydrogenase from Escher-

ichia coli, Biochemistry 36 (1997) 11331–11341.

[10] J. Rydstrom, Assay of nicotinamide nucleotide transhydrogenases in

mammalian, bacterial and reconstituted systems, Methods Enzymol.

55 (1979) 261–275.

[11] C. Johansson, A. Pedersen, B.G. Karlsson, J. Rydstrom, Redox-sen-

sitive loops D and E regulate NADP(H) binding in domain III and

domain I–domain III interactions in proton-translocating Escherichia

coli transhydrogenase, Eur. J. Biochem. 269 (2002) 4505–4515.

[12] J. Rydstrom, Site-specific inhibitors of mitochondrial nicotinamide

nucleotide transhydrogenase, Eur. J. Biochem. 31 (1972) 496–504.

[13] O. Fjellstrom, T. Bizouarn, J. Zhang, J. Rydstrom, J.D. Venning, J.B.

Jackson, Catalytic properties of hybrid complexes of the NAD(H)-

binding and NADP(H)-binding domains of the proton-translocating

transhydrogenases from Escherichia coli and Rhodospirillum rubrum,

Biochemistry 38 (1999) 415–422.

[14] O. Fjellstrom, M. Axelsson, T. Bizouarn, X. Hu, C. Johansson, J.

Meuller, J. Rydstrom, Mapping residues in the NADP(H)-binding site

of proton-translocating nicotinamide nucleotide transhydrogenase

from Escherichia coli. A study of structure and function, J. Biol.

Chem. 274 (1999) 6350–6359.

[15] J. Meuller, X. Hu, C. Bunthof, T. Olausson, J. Rydstrom, Identifica-

tion of an aspartic acid residue in the beta subunit which is essential

for catalysis and proton pumping by transhydrogenase from Escher-

ichia coli, Biochim. Biophys. Acta 1273 (1996) 191–194.

[16] J. Zhang, X. Hu, A.M. Osman, J. Rydstrom, Effects of metal ions on

the substrate-specificity and activity of proton-pumping nicotinamide

nucleotide transhydrogenase from Escherichia coli, Biochim. Bio-

phys. Acta 1319 (1997) 331–339.

[17] J. Meuller, J. Zhang, C. Hou, P.D. Bragg, J. Rydstrom, Properties of a

cysteine-free proton-pumping nicotinamide nucleotide transhydroge-

nase, Biochem. J. 324 (1997) 681–687.

[18] M. Althage, T. Bizouarn, J. Rydstrom, Identification of a region in-

volved in the communication between the NADP(H)-binding domain

and the membrane domain in proton-pumping E. coli transhydroge-

nase, Biochemistry 40 (2001) 9968–9976.

[19] R. Koradi, M. Billeter, K. Wutrich, MOLMOL: a program for display

and analysis of macromolecular structures, J. Mol. Graph. 14 (1996)

29–32.